A Method for Reducing Training Time of ML-Based Cascade Scheme for Large-Volume Data Analysis.

We live in the era of large data analysis, where processing vast datasets has become essential for uncovering valuable insights across various domains of our lives. Machine learning (ML) algorithms offer powerful tools for processing and analyzing this abundance of information. However, the considerable time and computational resources needed for training ML models pose significant challenges, especially within cascade schemes, due to the iterative nature of training algorithms, the complexity of feature extraction and transformation processes, and the large sizes of the datasets involved. This paper proposes a modification to the existing ML-based cascade scheme for analyzing large biomedical datasets by incorporating principal component analysis (PCA) at each level of the cascade. We selected the number of principal components to replace the initial inputs so that it ensured 95% variance retention. Furthermore, we enhanced the training and application algorithms and demonstrated the effectiveness of the modified cascade scheme through comparative analysis, which showcased a significant reduction in training time while improving the generalization properties of the method and the accuracy of the large data analysis. The improved enhanced generalization properties of the scheme stemmed from the reduction in nonsignificant independent attributes in the dataset, which further enhanced its performance in intelligent large data analysis.

Homogeneous Adaboost Ensemble Machine Learning Algorithms with Reduced Entropy on Balanced Data.

Today's world faces a serious public health problem with cancer. One type of cancer that begins in the breast and spreads to other body areas is breast cancer (BC). Breast cancer is one of the most prevalent cancers that claim the lives of women. It is also becoming clearer that most cases of breast cancer are already advanced when they are brought to the doctor's attention by the patient. The patient may have the evident lesion removed, but the seeds have reached an advanced stage of development or the body's ability to resist them has weakened considerably, rendering them ineffective. Although it is still much more common in more developed nations, it is also quickly spreading to less developed countries. The motivation behind this study is to use an ensemble method for the prediction of BC, as an ensemble model aims to automatically manage the strengths and weaknesses of each of its separate models, resulting in the best decision being made overall. The main objective of this paper is to predict and classify breast cancer using Adaboost ensemble techniques. The weighted entropy is computed for the target column. Taking each attribute's weights results in the weighted entropy. Each class's likelihood is represented by the weights. The amount of information gained increases with a decrease in entropy. Both individual and homogeneous ensemble classifiers, created by mixing Adaboost with different single classifiers, have been used in this work. In order to deal with the class imbalance issue as well as noise, the synthetic minority over-sampling technique (SMOTE) was used as part of the data mining pre-processing. The suggested approach uses a decision tree (DT) and naive Bayes (NB), with Adaboost ensemble techniques. The experimental findings shown 97.95% accuracy for prediction using the Adaboost-random forest classifier.

An Approach towards Increasing Prediction Accuracy for the Recovery of Missing IoT Data Based on the GRNN-SGTM Ensemble.

The purpose of this paper is to improve the accuracy of solving prediction tasks of the missing IoT data recovery. To achieve this, the authors have developed a new ensemble of neural network tools. It consists of two successive General Regression Neural Network (GRNN) networks and one neural-like structure of the Successive Geometric Transformation Model (SGTM). The principle of ensemble topology construction on two successively connected general regression neural networks, supplemented with an SGTM neural-like structure, is mathematically substantiated, which improves the accuracy of prediction results. The effectiveness of the method is based on the replacement of the summation of the results of the two GRNNs with a weighted summation, which improves the accuracy of the ensemble operation in general. A detailed algorithmic implementation of the ensemble method as well as a flowchart of its operation is presented. The parameters of the ensemble operation are determined by optimization using the brute-force method. Based on the developed ensemble method, the solution of the task of completing the partially missing values ​​in the real monitoring dataset of the air environment collected by the IoT device is presented. By comparing the performance of the developed ensemble with the existing methods, the highest accuracy of its performance (by the parameters of Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE) accuracy) among the most similar in this class has been proved.

An interpretable ensemble structure with a non-iterative training algorithm to improve the predictive accuracy of healthcare data analysis.

The modern development of healthcare is characterized by a set of large volumes of tabular data for monitoring and diagnosing the patient's condition. In addition, modern methods of data engineering allow the synthesizing of a large number of features from an image or signals, which are presented in tabular form. The possibility of high-precision and high-speed processing of such large volumes of medical data requires the use of artificial intelligence tools. A linear machine learning model cannot accurately analyze such data, and traditional bagging, boosting, or stacking ensembles typically require significant computing power and time to implement. In this paper, the authors proposed a method for the analysis of large sets of medical data, based on a designed linear ensemble method with a non-iterative learning algorithm. The basic node of the new ensemble is an extended-input SGTM neural-like structure, which provides high-speed data processing at each level of the ensemble. Increasing prediction accuracy is ensured by dividing the large dataset into parts, the analysis of which is carried out in each node of the ensemble structure and taking into account the output signal from the previous level of the ensemble as an additional attribute on the next one. Such a design of a new ensemble structure provides both a significant increase in the prediction accuracy for large sets of medical data analysis and a significant reduction in the duration of the training procedure. Experimental studies on a large medical dataset, as well as a comparison with existing machine learning methods, confirmed the high efficiency of using the developed ensemble structure when solving the prediction task.

Enhanced ANN-based ensemble method for bridge damage characterization using limited dataset.

Bridges are vital assets of transport infrastructure, systems, and communities. Damage characterization is critical in ensuring safety and planning adaptation measures. Nondestructive methods offer an efficient means towards assessing the condition of bridges, without causing harm or disruption to transport services, and these can deploy measurable evidence of bridge deterioration, e.g., deflections due to tendon loss. This paper presents an enhanced input-doubling technique and the Artificial Neural Network (ANN)-based cascade ensemble method for bridge damage state identification and is exclusively relying on small datasets, that are common in structural assessments. A new data augmentation scheme rooted in the principles of linearizing response surfaces is introduced, which significantly boosts the efficiency of intelligent data analysis when faced with limited volumes of data. Furthermore, improvements to a two-step ANN-based ensemble method, designed for solving the stated task, are presented. By adding the improved input-doubling methods as simple predictors in the first part of the cascade ensemble and optimizing it, we significantly boost accuracy (7%, 0.5%, and 8% based on R2 in predicting tendon losses for three critical zones that were defined across the deck of a real deteriorated prestressed balanced cantilever bridge). This improvement is strong evidence of the accuracy of the proposed method for the task at hand that is proven to be more accurate than other methods available in the international literature.

Smart systems and data-driven services in healthcare.

The modern development of Medicine and Healthcare is primarily based on the automation of various processes to support the correct and timely medical decisions. A doctor or other medical staff's fast, accurate, reliable diagnosis, prevention, and treatment processes are the keys to quality patient care. This aim can be achieved by developing reliable, intelligent Healthcare and medicine service systems for various purposes. The Special Issue covers science-intensive real-time solutions that underlie smart systems and data-driven services in Healthcare and Medicine.

An Approach toward Automatic Specifics Diagnosis of Breast Cancer Based on an Immunohistochemical Image.

The paper explored the problem of automatic diagnosis based on immunohistochemical image analysis. The issue of automated diagnosis is a preliminary and advisory statement for a diagnostician. The authors studied breast cancer histological and immunohistochemical images using the following biomarkers progesterone, estrogen, oncoprotein, and a cell proliferation biomarker. The authors developed a breast cancer diagnosis method based on immunohistochemical image analysis. The proposed method consists of algorithms for image preprocessing, segmentation, and the determination of informative indicators (relative area and intensity of cells) and an algorithm for determining the molecular genetic breast cancer subtype. An adaptive algorithm for image preprocessing was developed to improve the quality of the images. It includes median filtering and image brightness equalization techniques. In addition, the authors developed a software module part of the HIAMS software package based on the Java programming language and the OpenCV computer vision library. Four molecular genetic breast cancer subtypes could be identified using this solution: subtype Luminal A, subtype Luminal B, subtype HER2/neu amplified, and basalt-like subtype. The developed algorithm for the quantitative characteristics of the immunohistochemical images showed sufficient accuracy in determining the cancer subtype "Luminal A". It was experimentally established that the relative area of the nuclei of cells covered with biomarkers of progesterone, estrogen, and oncoprotein was more than 85%. The given approach allows for automating and accelerating the process of diagnosis. Developed algorithms for calculating the quantitative characteristics of cells on immunohistochemical images can increase the accuracy of diagnosis.

Cyber attack evaluation dataset for deep packet inspection and analysis.

To determine the effectiveness of any defense mechanism, there is a need for comprehensive real-time network data that solely references various attack scenarios based on older software versions or unprotected ports, and so on. This presented dataset has entire network data at the time of several cyber attacks to enable experimentation on challenges based on implementing defense mechanisms on a larger scale. For collecting the data, we captured the network traffic of configured virtual machines using Wireshark and tcpdump. To analyze the impact of several cyber attack scenarios, this dataset presents a set of ten computers connected to Router1 on VLAN1 in a Docker Bridge network, that try and exploit each other. It includes browsing the web and downloading foreign packages including malicious ones. Also, services like File Transfer Protocol (FTP) and Secure Shell (SSH) were exploited using several attack mechanisms. The presented dataset shows the importance of updating and patching systems to protect themselves to a greater extent, by following attack tactics on older versions of packages as compared to the newer and updated ones. This dataset also includes an Apache Server hosted on a different subset of VLAN2 which is connected to the VLAN1 to demonstrate isolation and cross- VLAN communication. The services on this web server were also exploited by the previously stated ten computers. The attack types include Distributed Denial of Service, SQL Injection, Account Takeover, Service Exploitation (SSH, FTP), DNS and ARP Spoofing, Scanning and Firewall Searching and Indexing (using Nmap), Hammering the services to brute-force passwords and usernames, Malware attacks, Spoofing, and Man-in-the-Middle Attack. The attack scenarios also show various scanning mechanisms and the impact of Insider Threats on the entire network.

A non-linear SVR-based cascade model for improving prediction accuracy of biomedical data analysis.

Biomedical data analysis is essential in current diagnosis, treatment, and patient condition monitoring. The large volumes of data that characterize this area require simple but accurate and fast methods of intellectual analysis to improve the level of medical services. Existing machine learning (ML) methods require many resources (time, memory, energy) when processing large datasets. Or they demonstrate a level of accuracy that is insufficient for solving a specific application task. In this paper, we developed a new ensemble model of increased accuracy for solving approximation problems of large biomedical data sets. The model is based on cascading of the ML methods and response surface linearization principles. In addition, we used Ito decomposition as a means of nonlinearly expanding the inputs at each level of the model. As weak learners, Support Vector Regression (SVR) with linear kernel was used due to many significant advantages demonstrated by this method among the existing ones. The training and application procedures of the developed SVR-based cascade model are described, and a flow chart of its implementation is presented. The modeling was carried out on a real-world tabular set of biomedical data of a large volume. The task of predicting the heart rate of individuals was solved, which provides the possibility of determining the level of human stress, and is an essential indicator in various applied fields. The optimal parameters of the SVR-based cascade model operating were selected experimentally. The authors shown that the developed model provides more than 20 times higher accuracy (according to Mean Squared Error (MSE)), as well as a significant reduction in the duration of the training procedure compared to the existing method, which provided the highest accuracy of work among those considered.

Special issue: informatics & data-driven medicine-2021.

Modern medical diagnosis, treatment, or rehabilitation problems of the patient reach completely different levels due to the rapid development of artificial intelligence tools. Methods of machine learning and optimization based on the intersection of historical data of various volumes provide significant support to physicians in the form of accurate and fast solutions of automated diagnostic systems. It significantly improves the quality of medical services. This special issue deals with the problems of medical diagnosis and prognosis in the case of short datasets. The problem is not new, but existing machine learning methods do not always demonstrate the adequacy of prediction or classification models, especially in the case of limited data to implement the training procedures. That is why the improvement of existing and development of new artificial intelligence tools that will be able to solve it effectively is an urgent task. The special issue contains the latest achievements in medical diagnostics based on the processing of small numerical and image-based datasets. Described methods have a strong theoretical basis, and numerous experimental studies confirm the high efficiency of their application in various applied fields of Medicine.

The functional safety assessment of cyber-physical system operation process described by Markov chain.

The functional safety assessment is one of the primary tasks both at the design stage and at the stage of operation of critical infrastructure at all levels. The article's main contribution is the information technology of calculating the author's metrics of functional safety for estimating the instance of the model of the cyber-physical system operation. The calculation of metric criteria analytically summarizes the results of expert evaluation of the system in VPR-metrics and the results of statistical processing of information on the system's operation presented in the parametric space Markov model of this process. The advantages of the proposed approach are the following: the need to process orders of magnitude less empirical data to obtain objective estimates of the investigated system; taking into account the configuration scheme and architecture of the security subsystem of the investigated system when calculating the metric; completeness, compactness, and simplicity of interpretation of evaluation results; the ability to assess the achievability of the limit values of the metric criteria based on the model of operation of the investigated system. The paper demonstrates the application of the proposed technology to assess the functional safety of the model of a real cyber-physical system.

Reliability model of the security subsystem countering to the impact of typed cyber-physical attacks.

The article's main contribution is the description of the process of the security subsystem countering the impact of typed cyber-physical attacks as a model of end states in continuous time. The input parameters of the model are the flow intensities of typed cyber-physical attacks, the flow intensities of possible cyber-immune reactions, and the set of probabilities of neutralization of cyber-physical attacks. The set of admissible states of the info-communication system is described taking into account possible variants of the development of the modeled process. The initial parameters of the model are the probabilities of the studied system in the appropriate states at a particular moment. The dynamics of the info-communication system's life cycle are embodied in the form of a matrix of transient probabilities. The mentioned matrix connects the initial parameters in the form of a system of Chapman's equations. The article presents a computationally efficient concept based on Gershgorin's theorems to solve such a system of equations with given initiating values. Based on the presented scientific results, the article proposes the concept of calculating the time to failure as an indicator of the reliability of the info-communication system operating under the probable impact of typical cyber-physical attacks. The adequacy of the model and concepts presented in the article is proved by comparing a statically representative amount of empirical and simulated data. We emphasize that the main contribution of the research is the description of the process of the security subsystem countering the impact of typed cyber-physical attacks as a model of end states in continuous time. Based on the created model, the concept of computationally efficient solution of Chapman's equation system based on Gershgorin's theorems and calculating time to failure as an indicator of the reliability of the info-communication system operating under the probable impact of typed cyber-physical attacks are formalized. These models and concepts are the highlights of the research.

CNGOD-An improved convolution neural network with grasshopper optimization for detection of COVID-19.

The world is facing the pandemic situation due to a beta corona virus named Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). The disease caused by this virus known as Corona Virus Disease 2019 (COVID-19) has affected the entire world. The current diagnosis methods are laboratory based and require specialized testing kits for performing the test. Therefore, to overcome the limitations of testing kits a diagnosis method from chest X-ray images is proposed in this paper. Chest X-ray images can be easily obtained by X-ray machines that are readily available at medical centres. The radiological examinations augmented with chest X-ray images is an effective way of disease diagnosis. The automated analysis of the chest X-ray images requires a highly efficient method for identifying COVID-19 from these images. Thus, a novel deep convolution neural network (CNN) optimized using Grasshopper Optimization Algorithm (GOA) is proposed. The deep learning model comprises depth wise separable convolutions that independently look at cross channel and spatial correlations. The optimization of deep learning models is a complex task due the multiple layers and their non-linearities. In image classification problems optimizers like Adam, SGD etc. get stuck in local minima. Thus, in this paper a metaheuristic optimization algorithm is used to optimize the network. Grasshoper Optimization Algorithm (GOA) is a metaheuristic algorithm that mimics the behaviour of grasshoppers for food search. This algorithm is a fast converging and is capable of exploration and exploitation of large search spaces. Maximum Probability Based Cross Entropy Loss (MPCE) loss function is used as it minimizes the back propogation error of cross entropy and improves the training. The experimental results show that the proposed method gives high classification accuracy. The interpretation of results is augmented with class activation maps. Grad-CAM visualization algorithm is used for class activation maps.

Modeling a session of subject-system interaction in a wireless communication infrastructure with a mixed resource.

The article examines the subject-system interaction session, where the system is understood as a base station, and the subject is understood as a mobile communication device. The peculiarity of the study is taking into account the phenomenon relevant to modern communication infrastructures, which is that the base station supports the division of information traffic into a subspace of guaranteed personalized traffic and a subspace of general-purpose traffic. The study considers a highly critical empirical emergency when the general-purpose traffic subspace may cease to be available at any time. The presented mathematical apparatus describes the impact of such an emergency on the active communication sessions supported by the system in receiving new incoming requests of increasing intensity. To characterize this emergency situation, expressions adapted for practical application are presented to calculate such qualitative parameters as the probability of stability, the probability of failure, and unavailability.

Mathematical modelling of bio-inspired frog leap optimization algorithm for transmission expansion planning.

Bio-inspired computing has progressed so far to deal with real-time multi-objective optimization problems. The Transmission expansion planning of the modern electricity grid requires finding the best and optimal routes for electricity transmission from the generation point to the endpoint while satisfying all the power and load constraints. Further, the transmission expansion cost allocation becomes a critical and pragmatic issue in the deregulated electricity industry. The prime objective is to minimize the total investment and expansion costs while considering N-1 contingency. The most optimal transmission expansion planning problem's solution is calculated using the objective function and the constraints. This optimal solution provides the total number and best locations for the candidates. The presented paper details the mathematical modeling of the shuffled frog leap algorithm with various modifications applied to the method to refine the results and finally proposes an enhanced novel approach to solve the transmission expansion planning problem. The proposed algorithm produces the expansion plans based on target-based evolution. The presented algorithm is rigorously tested on the standard Garver dataset and IEEE 24 bus system. The empirical results of the proposed algorithm led to better expansion plans while effectively considering typical electrical constraints along with modern and realistic constraints.

EDLDR: An Ensemble Deep Learning Technique for Detection and Classification of Diabetic Retinopathy.

Diabetic retinopathy (DR) is an ophthalmological disease that causes damage in the blood vessels of the eye. DR causes clotting, lesions or haemorrhage in the light-sensitive region of the retina. Person suffering from DR face loss of vision due to the formation of exudates or lesions in the retina. The detection of DR is critical to the successful treatment of patients suffering from DR. The retinal fundus images may be used for the detection of abnormalities leading to DR. In this paper, an automated ensemble deep learning model is proposed for the detection and classification of DR. The ensembling of a deep learning model enables better predictions and achieves better performance than any single contributing model. Two deep learning models, namely modified DenseNet101 and ResNeXt, are ensembled for the detection of diabetic retinopathy. The ResNeXt model is an improvement over the existing ResNet models. The model includes a shortcut from the previous block to next block, stacking layers and adapting split-transform-merge strategy. The model has a cardinality parameter that specifies the number of transformations. The DenseNet model gives better feature use efficiency as the dense blocks perform concatenation. The ensembling of these two models is performed using normalization over the classes followed by maximum a posteriori over the class outputs to compute the final class label. The experiments are conducted on two datasets APTOS19 and DIARETDB1. The classifications are carried out for both two classes and five classes. The images are pre-processed using CLAHE method for histogram equalization. The dataset has a high-class imbalance and the images of the non-proliferative type are very low, therefore, GAN-based augmentation technique is used for data augmentation. The results obtained from the proposed method are compared with other existing methods. The comparison shows that the proposed method has higher accuracy, precision and recall for both two classes and five classes. The proposed method has an accuracy of 86.08 for five classes and 96.98% for two classes. The precision and recall for two classes are 0.97. For five classes also, the precision and recall are high, i.e., 0.76 and 0.82, respectively.

A Deep Intelligent Attack Detection Framework for Fog-Based IoT Systems.

Fog computing provides a multitude of end-based IoT system services. End IoT devices exchange information with fog nodes and the cloud to handle client undertakings. During the process of data collection between the layer of fog and the cloud, there are more chances of crucial attacks or assaults like DDoS and many more security attacks being compromised by IoT end devices. These network (NW) threats must be spotted early. Deep learning (DL) assumes an unmistakable part in foreseeing the end client behavior by extricating highlights and grouping the foe in the network. Yet, because of IoT devices' compelled nature in calculation and storage spaces, DL cannot be managed on those. Here, a framework for fog-based attack detection is proffered, and different attacks are prognosticated utilizing long short-term memory (LSTM). The end IoT gadget behaviour can be prognosticated by installing a trained LSTMDL model at the fog node computation module. The simulations are performed using Python by comparing LSTMDL model with deep neural multilayer perceptron (DNMLP), bidirectional LSTM (Bi-LSTM), gated recurrent units (GRU), hybrid ensemble model (HEM), and hybrid deep learning model (CNN + LSTM) comprising convolutional neural network (CNN) and LSTM on DDoS-SDN (Mendeley Dataset), NSLKDD, UNSW-NB15, and IoTID20 datasets. To evaluate the performance of the binary classifier, metrics like accuracy, precision, recall, f1-score, and ROC-AUC curves are considered on these datasets. The LSTMDL model shows outperforming nature in binary classification with 99.70%, 99.12%, 94.11%, and 99.88% performance accuracies on experimentation with respective datasets. The network simulation further shows how different DL models present fog layer communication behaviour detection time (CBDT). DNMLP detects communication behaviour (CB) faster than other models, but LSTMDL predicts assaults better.

Special issue: Informatics & data-driven medicine.

The current state of the development of Medicine today is changing dramatically. Previously, data of the patient's health were collected only during a visit to the clinic. These were small chunks of information obtained from observations or experimental studies by clinicians, and were recorded on paper or in small electronic files. The advances in computer power development, hardware and software tools and consequently design an emergence of miniature smart devices for various purposes (flexible electronic devices, medical tattoos, stick-on sensors, biochips etc.) can monitor various vital signs of patients in real time and collect such data comprehensively. There is a steady growth of such technologies in various fields of medicine for disease prevention, diagnosis, and therapy. Due to this, clinicians began to face similar problems as data scientists. They need to perform many different tasks, which are based on a huge amount of data, in some cases with incompleteness and uncertainty and in most others with complex, non-obvious connections between them and different for each individual patient (observation) as well as a lack of time to solve them effectively. These factors significantly decrease the quality of decision making, which usually affects the effectiveness of diagnosis or therapy. That is why the new concept in Medicine, widely known as Data-Driven Medicine, arises nowadays. This approach, which based on IoT and Artificial Intelligence, provide possibilities for efficiently process of the huge amounts of data of various types, stimulates new discoveries and provides the necessary integration and management of such information for enabling precision medical care. Such approach could create a new wave in health care. It will provide effective management of a huge amount of comprehensive information about the patient's condition; will increase the speed of clinician's expertise, and will maintain high accuracy analysis based on digital tools and machine learning. The combined use of different digital devices and artificial intelligence tools will provide an opportunity to deeply understand the disease, boost the accuracy and speed of its detection at early stages and improve the modes of diagnosis. Such invaluable information stimulates new ways to choose patient-oriented preventions and interventions for each individual case.

Predictive modeling based on small data in clinical medicine: RBF-based additive input-doubling method.

The paper considers the problem of handling short sets of medical data. Effectively solving this problem will provide the ability to solve numerous classification and regression tasks in case of limited data in health decision support systems. Many similar tasks arise in various fields of medicine. The authors improved the regression method of data analysis based on artificial neural networks by introducing additional elements into the formula for calculating the output signal of the existing RBF-based input-doubling method. This improvement provides averaging of the result, which is typical for ensemble methods, and allows compensating for the errors of different signs of the predicted values. These two advantages make it possible to significantly increase the accuracy of the methods of this class. It should be noted that the duration of the training algorithm of the advanced method remains the same as for existing method. Experimental modeling was performed using a real short medical data. The regression task in rheumatology was solved based on only 77 observations. The optimal parameters of the method, which provide the highest prediction accuracy based on MAE and RMSE, were selected experimentally. A comparison of its efficiency with other methods of this class has been performed. The highest accuracy of the proposed RBF-based additive input-doubling method among the considered ones is established. The method can be modified by using other nonlinear artificial intelligence tools to implement its training and application algorithms and such methods can be applied in various fields of medicine.